Occupancy Grid Mapping
Star
Discuss
Download
In this mapping method, the 2-D world to be mapped is divided into evenly spaced grids, each grid has one of the three possible states: free, occupied and unknown. The robot wandering around in the world and take measurements (e.g. lidar scan) along the way, to obtain the pair
, these pairs are used to generate the map. The map is represented with a matrix, with each element corresponding to the state of a grid in the world. At each timestamp, the belief of state of the grid is updated according to binary (free vs occupied) Bayes filter using the old belief and the measurement (inverse measurement model i.e.
) (grid with unknown state is left untouched).